Trying to figure out if there is something in my configuration that prevents some nodes , edge devices that have full connectivity , they are visible in the visualization map but have no connection lines to them. Sometimes they might appear with a connection line if i completely remove them and reconnect them. I.e. this could be related to config or way of joining.
This happens in 2 of my HA instances , one that created connection lines after a few days , the other has been running without enabling connections for a few weeks.
As this is both temperature sensors and magnet sensors and same brand where some works with connection lines and some still works but without connection lines its a little confusing.
Logs indicates traffic from the working devices that looks the same even if i dont have knowledge on exact which messages to look for.
Are the devices still working or do they stop working? The lines disappearing is most likely because they are battery powered and therefore go to sleep. Look at the FAQ here (Z2M but same pricipal for ZHA). FAQ | Zigbee2MQTT
ZHA and Z2M have a time setting for when a device is unavailable.
If a device has not responded in this time interval, then it is considered unavailable.
The keyword here is “considered”, because devices decide themself how long they want to sleep and be unavailable.
The time setting can be changed so the devices are not listed as unavailable that often or not at all, but the more you increase the time the longer you are to discover devices that have actually gone unavailable for real.
Still working , A thing that can be noted is that in the HA log when enabling debugging all devices seems to update periodically. I.e. NWK is seen and the DB is updated with last seen as it should be. Normally when a device is getting “offline” the battery indicator in the device list is gone for offline devices
Not changed the standard offline timers. And i can see last seen to be recently updated. It is not either indicated as offline.
I have had issues with edge devices that i hopefully now have resolved with explicit repeaters that seems to have better radio performance than for example a outlet plug etc.
And what is recently updated in your perspective?
ZHA have a default 2 hours timeout for mains powered device and 6 hours for a battery powered device.
What might be strange and im not really sure why this gets more frequent except if its the debug mode that triggers more traffic in ZHA.
Got 1 alive message every minute…
All nodes seems to have the same timing when looking at the log.
I.e. grep for TH01 (temp) or DS01(magnet)
I.e. rather frequent. Coverage… its about 4 meter from the controller and do have routers rather close to it even if i have no clue on which one it selects if anything else than the controller.
On a TH01 that according to connectivity lines jump via a range extender the traffic is rather similar.
Unless then the debug mode in ZHA does more than just increase logging level.
Ill see if i can figure out what’s happening…
The device is reachable as its not getting a offline indication.
If the messages really are that frequent it seems not to affect the battery at least.
Still around 100% for most devices except the temp i put in the freezer which has dropped to 55% since aug2023 something.
The ZHA “network visualisation” was originally a dashboard card. It was absorbed into ZHA and the card is now deprecated, but the repository is still on Github if you’re interested.
The lines on the map show every discovered route between nodes - not just the ones ZHA is using. They are colour coded according to their LQI. The map is a snapshot. It will be different each time you generate it.
It is normal to see nodes on the map that don’t have a link to anything else. These are usually end devices and they werere probably sleeping to conserve battery at the moment the map was generated. Connections to end devices are initiated by the end devices themselves - parent routers can’t “ping” them to see if they’re working. They will wake immediately if one of their sensors is triggered, and in any event they should wake periodically to check in with their parent router.
Not all sensor values are updated at the same time. They will update if there is a change, but if there isn’t a change frequency will vary with the manufacturer - this is part of a planned strategy to extend battery life.
This is a Hue motion sensor, where light level and temperature are changing, but there has been no movement for some time. If there is no change, battery level is updated (not very accurately) once every 24 hours:
An update to any one of those will flag the device as “available”. If none of them is updated the device will eventually time out and be marked as “unavailable” (as @WallyR says, default timeout is 6 hours on ZHA).
If no sensors are activated a “sleeping” device will wake periodically to send a “check in” update, but otherwise it will appear to have no connection on the map. You could easily generate a map several times over several days and find it sleeping each time - that doesn’t mean it hasn’t woken up in between times, and it will be marked as “available” because of that.
Nice!
Ill have a look at the repository and see if i can figure out why this “single” node is not having any connectivity lines. Even a complete reboot of the RPI that im running on changes anything. I.e. connectivity links seems to stay and the one missing is still missing.
If i understand correct then a connectivity link can appear or disappear over time and is based on a snapshot in time or what has been seen perhaps for the last hour or 2 related to communication from a specific node.
One explicit snapshot during a few seconds would probably have no or very few connectivity lines… And as we can see older / unused paths its probably based on what has been seen over time.
Update. Had a look on ZHA_MAP ( the old repository ) to try to find out if something there could cause the lost links , specifically between EDGE and the rest of the network.
Seems like people have noticed this earlier and there is a note about missing neighbor relations in zigbee.db.
That could be confirmed… No neighbor relations for the device with a missing link.
Someone mentions that a edge might not bee connected correct if joined via the controller , if joined via a router it works.
On a different HA installation i today replaced a RPI4 with a RPI5 and the same happened. I.e. EDGE with no link. Interesting this device can report data into the network. I assume it them have learned keys and can send understandable data in the network. But not visible in zigbee.db, ( looking with a SQLite browser )
Other EDGE’s exist in multiple tables in zigbee.db but not this one that have no connectivity link.
I would then assume that its really a ZHA issue and not a ZHA_MAP ( even if this now is incorporated in ZHA ).
For ZHA… I guess this happens during initial registration of a ZigBee device.
Ill dig some more with debug log enabled and si if there is anything obvious.
